Much is told about gender pay gap and the actions needed (and taken) to overcome it.
But have we managed to defeat it? Are we paying women less than men? And if so, are we mitigating that inequality across time?
The Gender Pay Gap Service in the UK is concentrating relevant data since 2017, giving us the oppotrunity to seek answers to such questions.
The current report seeks to deploy them and provide insights.
Data is downloaded from the Official Website of the Gender Pay Gap Service and span across the period 2017 - 2021.
After parsing all relevant CSVs, our dataframe more or less looks like:
## 'data.frame': 32631 obs. of 16 variables:
## $ EmployerName : chr "\"Bryanston School\",Incorporated" "\"RED BAND\" CHEMICAL COMPANY, LIMITED" "118 LIMITED" "123 EMPLOYEES LTD" ...
## $ PostCode : chr "DT11 0PX" "EH6 8NU" "CF24 2SA" "LS7 1AB" ...
## $ Sic_Code : int 85310 47730 61900 78300 93110 56210 93110 86900 56290 1470 ...
## $ Description : chr "General secondary education" "Dispensing chemist in specialised stores" "Other telecommunications activities" "Human resources provision and management of human resources functions" ...
## $ DiffMeanHourlyPercent : num 18 2.3 1.7 41 -22 13.4 15.1 15 11.9 13.4 ...
## $ DiffMedianHourlyPercent: num 28.2 -2.7 2.8 36 -34 8.1 2.8 0 0 8.5 ...
## $ DiffMeanBonusPercent : num 0 15 13.1 -69.8 -47 41.4 77.6 0 0 62.9 ...
## $ DiffMedianBonusPercent : num 0 37.5 13.6 -157.2 -67 ...
## $ MaleBonusPercent : num 0 15.6 70 50 25 8.7 5.8 0 0 89.7 ...
## $ FemaleBonusPercent : num 0 66.7 57 73.5 75 3.2 4.2 0 0 73.3 ...
## $ MaleLowerQuartile : num 24.4 20.3 51 0 56 29.1 42.6 10 75.3 89.1 ...
## $ FemaleLowerQuartile : num 75.6 79.7 49 100 44 70.9 57.4 90 24.7 10.9 ...
## $ MaleTopQuartile : num 51.5 18.1 58 23 24 58.2 35.5 9 69.8 93.8 ...
## $ FemaleTopQuartile : num 48.5 81.9 42 77 76 41.8 64.5 91 30.2 6.2 ...
## $ EmployerSize : chr "500 to 999" "250 to 499" "500 to 999" "250 to 499" ...
## $ Year : int 2017 2017 2017 2017 2017 2017 2017 2017 2017 2017 ...Where:
The Mean & Median Pay Columns (DiffMeanHourlyPercent, DiffMedianHourlyPercent, etc.) stand for the Mean and Median % difference between male and female hourly pay (negative = women’s mean hourly pay is higher). Bear in mind that whenever these figures are positive, they signify a larger pay on men. So, for example a company that reported a DiffMeanHourlyPercent of 20, pays men 20% more than pays women.
(Fe)MaleBonusPercent indicates the Percentage of (fe)male employees paid a bonus.
Columns containing Quartile (MaleLowerQuartile, etc.) represent the percentage that each corporate level is occupied by each gender. MaleLowerQuartile + FemaleLowerQuartile (and so on for the rest of such columns) always add up to 1.
The rest of the Columns are self-explanatory.
Here you can find the official descriptions published by the respective service.
Let’s start exploring our data.
For starters, out of all companies, how many favor men and how many women?
At first glance, the vast majority of companies in the UK favor men over women in pay.
When considering the Median Difference in pay, inequalities appear to smooth a bit. One possible explanation for that is that the highest paid employees are men. This fact forces the mean difference towards the men’s side.
But we will examine it further later in this report. For now, we will go on with the Mean instead of the Median Difference because we want to take into account such inequalities derived from extremely high pays.
Aside from that, a slight improvement seems to come into sight for 2021. But since this report is produced in mid-2021, many companies haven’t reported data yet.
A glance into the absolute numbers of companies favoring a gender might enlighten us:
And indeed, data for 2021 is very scarce, just 231.
That’s probably the reason why a positive change appeared in the earlier plot.
So far, we just counted each difference no matter its magnitude. Even just a 0.01% difference is counted as favoring men (and a -0.01% for women respectively).
So, let’s take a look at the actual numbers of differences to enlighten their magnitude.
To interpret a Boxplot (like the preceding plot) it’s easier to consider an example of an 100 observations population.
In such an example:
the lower boundary of each box represents the 25th highest difference
the line inside each box represents the 50th highest difference
the upper boundary of each box represents the 75th highest difference
dots represent outliers.
That being said, women’s distribution differences (i.e. all negative differences found in the dataset) is ridiculusly closer to zero than the respective men’s distribution of differences.
In addition, men’s lower 25% starts where women’s 75% ends, not to mention their median difference (15.1333333% for men against a mere 3.65% for women).
Overall, it seems that when a company favors women in pay, the difference is small and close to 0%, while when men are paid more, their difference from women is more significant.
Apart from ordinary pay, many companies compensate their employees with bonuses.
As stated earlier, our dataset contains information about the percentage of the total (fe)male employees that received a bonus.
Let’s take a look at the distribution of these percentages:
The two genders’ bonus percentages hopefully follow very similar distributions.
For the most part, the two distributions intersect.
The only parts where the differentiate are:
at the very start (from 0% to 3%) where once again women experience many more 0%s of bonus than men
at the very end, where men receive more 84%s to 95%s than do women.
Apart from the percentage, what’s going on with the differences in bonus pay?
As with the ordinary pay section, we will explore:
the percentage of total companies that favor one gender in bonus pay
the magnitudes of those favors
Very similarly with the ordinary pay,
most companies pay greater bonuses to men than to women
when women receive a greater bonnus, the difference with men’s bonus is usually close to zero, in contrast with men
Let’s now explore the rankings structure of the UK companies.
Our dataset contains information about the composition of their: * Lower * Lower Middle * Upper Middle and * Top level executives.
As an example, if a company’s lower level is consisted of 70% women, the men’s respective percentage will be 30% and they always add up to 1.
Let’s first take a look at their composition regardless of the year:
It seems that women dominate lower and lower middle levels, while men prevail the top ones.
In other words, most companies report 75%s of women (and 25%s for men respectively) in their lower quartile and vice versa for their top quartiles.
That also explains the discrepancy shown in the first plot between the mean and median pay. Mean pay tends towards men because they hold the highest positions (and thus the highest salaries) in most companies.
Is this tendency weakened across time? Are things getting more diverse in the steps of the corporate ladder?
It appears that no.
The same pattern is observed every year. Men on top, women underneath them.
The only exception is year 2021, which, as we saw earlier, contains way too few observations to present a clear picture.
Could some of the inequalities we observed so far be attributed to some sectors (such as Science, Technology, Engineering, Mathematics) that pay well and are dominated by a gender?
So first, which are the top 10 sectors that pay better each gender?
No pattern can be observed regarding the nature of those sectors.
But what can be observed is yet again the magnitude of differences.
The sector that pays women best, Window cleaning services, is paying women just 28.6 % more than is paying men.
On the other hand, the 10th sector that pays men best, Support activities for crop production, is paying better even than women’s best, at 34.4%.
Additionaly, out of all sectors, what percentage is paying men better than women?
Yet again, most sectors pay men better than women.
Apart from the sectors, could regions contribute to inequalities?
First of all, in no region women are getting paid better than men.
Apart from that, no region seems to concentrate certain discrepancies (either low or high), at least at first glance.
One could just say that the South has more sharpened inequalities in comparison to the British North, but still needs further examination.
Could employer Size be a factor that determines the gender pay gap?
Let’s take a look at each size’s average difference of pay by gender:
Pretty much pay is biased towards men no matter the size of company.
Overall, women are paid less than men, both in terms of ordinary pay and bonus.
Upper level quartiles of the corporate ladder are dominated by men, while lower ones are dominated by women.
In the majority of sectors women are on average paid less than men.
In all regions and company sizes, women are on average paid less than men.